beeline audio Installation The audio setup employed three computers hosting systems that listened to each other and responded. Node 1 output left channel to Node 2, right channel to Node 3. Node 2 output left channel to Node 3, right channel to Node 1. Node 3 output left channel to Node 1, right channel to Node 2. Each node sampled incoming audio, chose a playback speed and filter frequency and Q slope to apply, and sent the transformed audio back to the other two nodes. All parameters were chosen at random from ranges derived from a simple 'melodic' sequence of integers. To create space, each node had a master clock that switched its activity on and off every 70 seconds or so. To ensure that the system would not subside into silence, a seed sample was used to feed audio in. The seed sample was violin music recorded from a speaker in a storefront alcove on Queen Street. The edited sample cut the melody line off in mid-phrase, left a silent space, and began again. The result was an apparent dialogue betwen the truncated violin melody and more or less corrosive responses from the three nodes. These reponses would build to flurries of activity and retreat to periods of silence. Performance For performance the automatic responses of two of the nodes were replaced with manual control over the timing of sample recording, playback speed, and filter gain, centre frequency, and Q. The third node was switched between periods of more and less dense response activity, and hosted an eight-channel spatializer that allowed four-channel panning patterns for each of the three nodes to be created during performance. Curves for extreme ramping changes in playback speed were employed. The seed samples were: strollers on the boardwalk at the beach on Queen St. East, a carpet auctioneer, and a dripping tap in our hotel room.